1. The transvascular fluid balance has been studied in isolated blood-perfused lungs of rabbit and cat. During periods of left atrial pressure (LAp) elevation, changes in preparation weight were measured. In some experiments simultaneous recordings of changes in blood volume, as judged by changes in emitted radioactivity from (51)Cr labelled red blood cells, were performed.2. In nine out of eleven experiments, elevation of LAp by 5 mm Hg within a test period of 8 min resulted in new weight stability and thus apparently in the establishment of a new transvascular fluid equilibrium. In five of these eleven experiments weight stability was also obtained on elevation of LAp by 10 mm Hg. When LAp was elevated further, transvascular fluid equilibrium was not obtained, and a continuous and eventually progressive outward flux of fluid occurred.3. In a preparation with the vasculature paralysed by papaverine in order to avoid vasomotor reactions, changes in plasma protein content caused immediate alterations in the weight slopes of the preparations. These alterations were interpreted as being due to changes in transvascular fluid transfer. Again new stable weights were obtained on moderate changes in plasma protein content.4. New weight and transvascular fluid equilibria seemed to be reached as a result of very small transvascular fluxes of fluid. The transvascular fluid balance in the lung may thus be dynamically maintained as a result of such small fluxes of fluid. The Starling hypothesis could account for this balance, provided that a small shift of fluid can cause sufficiently marked alterations in the osmotic and/or hydrostatic pressure gradient across the wall of the exchange vessels.5. The possible role of a small extravascular compartment with a high fluid colloid osmotic pressure is discussed in relation to transvascular fluid balance in the lung.